R.T. Cederwall

To assess the current status of climate models in simulating clouds, basic cloud climatologies from ten atmospheric general circulation models are compared with satellite measurements from the International Satellite Cloud Climatology Project (ISCCP) and the Clouds and Earth's Radiant Energy System (CERES) program. An ISCCP simulator is employed in all models to facilitate the comparison. Models simulated a four‐fold difference in high‐top clouds. There are also, however, large uncertainties in satellite high thin clouds to effectively constrain the models. The majority of models only simulated 30–40% of middle‐top clouds in the ISCCP and CERES data sets. Half of the models underestimated low clouds, while none overestimated them at a statistically significant level. When stratified in the optical thickness ranges, the majority of the models simulated optically thick clouds more than twice the satellite observations. Most models, however, underestimated optically intermediate and thin clouds. Compensations of these clouds biases are used to explain the simulated longwave and shortwave cloud radiative forcing at the top of the atmosphere. Seasonal sensitivities of clouds are also analyzed to compare with observations. Models are shown to simulate seasonal variations better for high clouds than for low clouds. Latitudinal distribution of the seasonal variations correlate with satellite measurements at >0.9, 0.6–0.9, and −0.2–0.7 levels for high, middle, and low clouds, respectively. The seasonal sensitivities of cloud types are found to strongly depend on the basic cloud climatology in the models. Models that systematically underestimate middle clouds also underestimate seasonal variations, while those that overestimate optically thick clouds also overestimate their seasonal sensitivities. Possible causes of the systematic cloud biases in the models are discussed.

Abstract Large‐eddy simulations of the development of shallow cumulus convection over land are presented. Many characteristics of the cumulus layer previously found in simulations of quasi‐steady convection over the sea are found to be reproduced in this more strongly forced, unsteady case. Furthermore, the results are shown to be encouragingly robust, with similar results obtained with eight independent models, and also across a range of numerical resolutions. The datasets produced are already being used in the development and evaluation of parametrizations used in numerical weather‐prediction and climate models. © Royal Meteorological Society, 2002. A. R. Brown's, A. P. Lock's and M. K. MacVean's contributions are Crown copyright.

Motivated by the need to obtain accurate objective analysis of field experimental data to force physical parameterizations in numerical models, this paper first reviews the existing objective analysis methods and interpolation schemes that are used to derive atmospheric wind divergence, vertical velocity, and advective tendencies. Advantages and disadvantages of different methods are discussed. It is shown that considerable uncertainties in the analyzed products can result from the use of different analysis. The paper then describes a hybrid approach to combine the strengths of the regular grid and the line-integral methods, together with a variational constraining procedure for the analysis of field experimental data. In addition to the use of upperair data, measurements at the surface and at the top of the atmosphere (TOA) are used to constrain the upperair analysis to conserve column-integrated mass, water, energy, and momentum.

To significantly improve the simulation of climate by general circulation models (GCMs), systematic errors in representations of relevant processes must first be identified, and then reduced. This endeavor demands that the GCM parameterizations of unresolved processes, in particular, should be tested over a wide range of time scales, not just in climate simulations. Thus, a numerical weather prediction (NWP) methodology for evaluating model parameterizations and gaining insights into their behavior may prove useful, provided that suitable adaptations are made for implementation in climate GCMs. This method entails the generation of short-range weather forecasts by a realistically initialized climate GCM, and the application of six hourly NWP analyses and observations of parameterized variables to evaluate these forecasts. The behavior of the parameterizations in such a weather-forecasting framework can provide insights on how these schemes might be improved, and modified parameterizations then can be tested in the same framework. To further this method for evaluating and analyzing parameterizations in climate GCMs, the U.S. Department of Energy is funding a joint venture of its Climate Change Prediction Program (CCPP) and Atmospheric Radiation Measurement (ARM) Program: the CCPP-ARM Parameterization Testbed (CAPT). This article elaborates the scientific rationale for CAPT, discusses technical aspects of its methodology, and presents examples of its implementation in a representative climate GCM.

Abstract This paper reports an intercomparison study of midlatitude continental cumulus convection simulated by eight two‐dimensional and two three‐dimensional cloud‐resolving models (CRMs), driven by observed large‐scale advective temperature and moisture tendencies, surface turbulent fluxes, and radiative‐heating profiles during three sub‐periods of the summer 1997 Intensive Observation Period of the US Department of Energy's Atmospheric Radiation Measurement (ARM) program. Each sub‐period includes two or three precipitation events of various intensities over a span of 4 or 5 days. The results can be summarized as follows. CRMs can reasonably simulate midlatitude continental summer convection observed at the ARM Cloud and Radiation Testbed site in terms of the intensity of convective activity, and the temperature and specific‐humidity evolution. Delayed occurrences of the initial precipitation events are a common feature for all three sub‐cases among the models. Cloud mass fluxes, condensate mixing ratios and hydrometeor fractions produced by all CRMs are similar. Some of the simulated cloud properties such as cloud liquid‐water path and hydrometeor fraction are rather similar to available observations. All CRMs produce large downdraught mass fluxes with magnitudes similar to those of updraughts, in contrast to CRM results for tropical convection. Some inter‐model differences in cloud properties are likely to be related to those in the parametrizations of microphysical processes. There is generally a good agreement between the CRMs and observations with CRMs being significantly better than single‐column models (SCMs), suggesting that current results are suitable for use in improving parametrizations in SCMs. However, improvements can still be made in the CRM simulations; these include the proper initialization of the CRMs and a more proper method of diagnosing cloud boundaries in model outputs for comparison with satellite and radar cloud observations. Copyright © 2002 Royal Meteorological Society